This component project of "Automated X-ray Crystallography for Structural Genomics" focuses on the construction of the computational foundation for the new PHENIX system will permit all tasks required for the computation of phases, automated map interpretation, model building and refinement, to be integrated with newly developed procedures for automated decision-making. Our ultimate goal is the creation of a system that will take X-RAY diffraction data and rapidly produce rapidly produce minimally biased atomic coordinates with little or no human intervention. I. Computational foundation for automated crystallography. Our new system for automated crystallography (PHENIX) will be built on strong foundation that can accommodate our current plans and future developments. We will develop a toolkit that will provide an environment for integrating crystallographic tasks and automated decision-making. The entire history of the structure. Along with all the structural information generated, will be stored in a Project History database (PHdb). II. Automated decision-making. We will develop a high-level framework for the expression of decision-making and crystallographic strategies. This will provide a transparent mechanism for structure solution strategies to be constructed using a graphical interface strategies. This will provide a transparent mechanism for structure solution strategies to be constructed using a graphical interface. The flexible nature of the decision-making framework will enable us to construct different views of the structure solution process depending on the needs of the user. The structural genomics user will be presented with a single button for structure solution, whereas the expert user will be able to request more detailed access to the structure solution tools. III. Heavy atom location. As part of the CNS package we have previously implemented a successful algorithm for heavy atom location that was based on Patterson methods. Currently the significant variability of experimental conditions requires the user to choose between a variety of available methods. We will combine our algorithm with those developed in Project II and Project III. The integrated nature of the PHENIX will allow information from later stages of structure solution to be used to improve the heavy atom model. This will yield more accurate experimental phases and an improved electron density map that will be more amenable to automated interpretation. IV. Integrated map interpretation and model refinement. We will automate structure completion by combining advanced optimization algorithms with state-of-the-art maximum likelihood refinement targets developed in Project III, and pattern-matching map interpretation methods developed in Project IV. In an automated structure determination system optimal parameterization of the model will be crucial. Analysis of the data quality followed by trials with various types of models will be used to determine which parameterization gives the best fit as judged by an objective criterion such as the free R-value.